graphembed-rs 0.1.0__cp310-cp310-macosx_11_0_arm64.whl → 0.1.2__cp310-cp310-macosx_11_0_arm64.whl

graphembed_rs/__init__.py

@@ -0,0 +1,7 @@
+from importlib import import_module as _imp
+from graphembed_rs import *      
+
+# add the helper into the same namespace
+from .load_utils import *                   
+
+del _imp    

graphembed_rs/load_utils.py

@@ -0,0 +1,71 @@
+import numpy as np
+from pathlib import Path
+### install pymongo with pip install pymongo
+# -- pymongo is a BSON decoder, not a BSON encoder --
+from bson import decode_file_iter
+
+
+def _dtype_from_type_name(tname: str):
+    """Map the Rust type-name string to a NumPy dtype."""
+    if tname == "f32":
+        return np.float32
+    if tname == "f64":
+        return np.float64
+    if tname == "usize":                 # stored as i64 in BSON
+        return np.int64
+    raise ValueError(f"Unknown type_name {tname!r} in BSON header")
+
+
+def load_embedding_bson(path: str | Path, *, want_in: bool = False):
+    """
+    Parameters
+    ----------
+    path : str | pathlib.Path
+        File written by graphembed::io::bson_dump(...)
+    want_in : bool, default False
+        • False … always return the OUT/source embedding (shape = (n, d))  
+        • True  … additionally return the IN/target embedding when
+          the dump is *asymmetric* (tuple(out, in_)).  For symmetric
+          dumps the second item is None.
+
+    Returns
+    -------
+    np.ndarray            (sym. dump or want_in=False)
+    OR (out_emb, in_emb)  (want_in=True)
+    """
+    path = Path(path)
+    with path.open("rb") as fh:
+        docs = decode_file_iter(fh)
+
+        # -- header ----------------------------------------------------
+        header = next(docs)["header"]
+        n      = int(header["nbdata"])
+        d      = int(header["dimension"])
+        sym    = bool(header["symetric"])
+        dtype  = _dtype_from_type_name(header["type_name"])
+        out_emb = np.empty((n, d), dtype=dtype)
+        in_emb  = None if sym or not want_in else np.empty((n, d), dtype=dtype)
+
+        # --OUT part --------------------------------------------------
+        for _ in range(n):
+            doc = next(docs)
+            key, vec = next(iter(doc.items()))            # only 1 (key,val)
+            idx, tag = map(int, key.split(","))
+            assert tag == 0, f"expected tag 0, got {tag}"
+            out_emb[idx] = np.asarray(vec, dtype=dtype)
+
+        # -- IN part (if any) -----------------------------------------
+        if not sym:
+            for _ in range(n):
+                doc = next(docs)
+                key, vec = next(iter(doc.items()))
+                idx, tag = map(int, key.split(","))
+                assert tag == 1, f"expected tag 1, got {tag}"
+                if in_emb is not None:                    # want_in == True
+                    in_emb[idx] = np.asarray(vec, dtype=dtype)
+                # else: silently drop it
+
+        # -- optional indexation doc – skip for now --------------------
+        # (decode_file_iter stops automatically at EOF)
+
+    return (out_emb, in_emb) if want_in else out_emb

{graphembed_rs-0.1.0.dist-info → graphembed_rs-0.1.2.dist-info}/METADATA

@@ -1,30 +1,101 @@
 Metadata-Version: 2.4
 Name: graphembed_rs
-Version: 0.1.0
+Version: 0.1.2
+Requires-Dist: numpy>=2.2.5
+Requires-Dist: pymongo>=4.12.1
+License-File: LICENSE-MIT
 Summary: Python bindings for the high‑performance Rust graph/network embedding library graphembed
 Keywords: graph,embedding,hash
 Author: Jianshu Zhao
-Author-email: jeanpierre.both@gmail.com, jianshuzhao@yahoo.com
+Author-email: jeanpierre.both@gmail.com
 License: MIT OR Apache-2.0
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
 Project-URL: Source Code, https://github.com/jean-pierreBoth/graphembed
 
-# Graphembed
-
-This crate provides ,as a library and an executable,embedding of directed or undirected graphs with positively weighted edges.
-
-
-  - For simple graphs, without data attached to nodes/labels, there are 2 (rust) modules **nodesketch** and **atp**. A simple executable with a validation option based on link prediction is also provided.
-
-  - To complement the embeddings we provide also core decomposition of graphs (see the module **structure**). We give try to analyze how Orkut communities are preserved through an embedding. (See Notebooks directory).
-
-  - The module **gkernel** is dedicated to graphs with discrete labels attached to nodes/edges. We use the *petgraph* crate for graph description.
-    The algorithm is based on an extension of the hashing strategy used in the module **nodesketch**.  
-    In the undirected case, this module also computes a global embedding vector for the whole graph. **It is still in an early version**.
+[![install with bioconda](https://img.shields.io/badge/install%20with-bioconda-brightgreen.svg?style=flat)](http://bioconda.github.io/recipes/graphembed/README.html)
+![](https://anaconda.org/bioconda/graphembed/badges/license.svg)
+![](https://anaconda.org/bioconda/graphembed/badges/version.svg)
+![](https://anaconda.org/bioconda/graphembed/badges/latest_release_relative_date.svg)
+![](https://anaconda.org/bioconda/graphembed/badges/platforms.svg)
+[![install with conda](https://anaconda.org/bioconda/graphembed/badges/downloads.svg)](https://anaconda.org/bioconda/graphembed)
+
+<div align="center">
+  <img width="35%" src ="GraphEmbed_log.jpg">
+</div>
+
+# GraphEmbed: Efficient and Robust Network Embedding via High-Order Proximity Preservation or Recursive Sketching
+
+This crate provides an executable and a library for embedding of directed or undirected graphs with positively weighted edges. We engineered and optimized current network embedding algorithms for large-scale network embedding, especially biological network. This crate was developed by [Jianshu Zhao](https://gitlab.com/Jianshu_Zhao) and Jean-Pierre Both [jpboth](https://gitlab.com/jpboth). We have a copy here in [Github](https://github.com/jianshu93/graphembed)
+
+
+  - For simple graphs, without data attached to nodes, there are 2 modules **nodesketch** and **atp**. A simple executable with a validation option based on link prediction is also provided.
+
+## Quick Install
+
+### Pre-built binaries on Linux
+```bash
+wget https://gitlab.com/-/project/64961144/uploads/9d7d0b038140cb67c584f01cd6dafac9/graphembed_Linux_x86-64_v0.1.6.zip
+unzip graphembed_Linux_x86-64_v0.1.6.zip
+chmod a+x ./graphembed
+./graphembed -h
+```
+
+### Bioconda on Linux/MacOS
+```bash
+conda install -c conda-forge -c bioconda graphembed
+```
+
+### Homebrew on MacOS
+```bash
+brew tap jianshu93/graphembed
+brew update
+brew install graphembed
+```
+
+
+### In Python (Please install python>=3.9 first)
+```bash
+pip install graphembed_rs
+
+### or you can build from source (Linux) after installing maturin
+git clone https://gitlab.com/Jianshu_Zhao/graphembed
+cd graphembed
+pip install maturin
+### note: for macOS, you need to change the line "features = ["pyo3/extension-module", "intel-mkl-static", "simdeez_f"]" in pyporject.toml to "features = ["pyo3/extension-module","openblas-system","stdsimd"]", you also need to Install OpenBLAS and add to system library path via Homebrew
+maturin develop --release
+
+#### Prepare some data
+wget https://gitlab.com/-/project/64961144/uploads/4e341383d62d86d1dd66e668e91b2c07/BlogCatalog.txt
+```
+
+```python
+import os
+os.environ["RUST_LOG"] = "info"
+import graphembed_rs.graphembed_rs as ge
+import graphembed_rs.load_utils as ge_utils
+help(ge)
+help(ge_utils)
+### HOPE
+ge.embed_hope_rank("BlogCatalog.txt", target_rank=128, nbiter=4,output="embedding_output")
+out_vectors=ge_utils.load_embedding_bson("embedding_output.bson")
+print("OUT embedding shape :", out_vectors.shape)
+print("first OUT vector    :", out_vectors[0])
+
+### Sketching
+### sketching only
+ge.embed_sketching("BlogCatalog.txt", decay=0.3, dim=128, nbiter=5, symetric=True, output="embedding_output")
+out_vectors=ge_utils.load_embedding_bson("embedding_output.bson")
+print("OUT embedding shape :", out_vectors.shape)
+print("first OUT vector    :", out_vectors[0])
+
+
+### validate accuracy
+auc_scores = ge.validate_sketching("BlogCatalog.txt",decay=0.3, dim=128, nbiter=3, nbpass=1, skip_frac=0.2,symetric=True, centric=True)
+print("Standard AUC per pass:", auc_scores)
+```
 
 ## Methods
-
 ### The embedding algorithms used in this crate are based on the following papers
 
 - **nodesketch**
@@ -57,29 +128,12 @@ Source node are related to left singular vectors and target nodes to the right o
 The similarity measure is the dot product, so it is not a norm.  
 The svd is approximated by randomization as described in Halko-Tropp 2011 as implemented in the [annembed crate](https://crates.io/crates/annembed).
 
-### The core decomposition algorithms
-
-- **Density-friendly decomposition**
-
-*Large Scale decomposition via convex programming  2017*  
-    M.Danisch T.H Hubert Chan and M.Sozio
-
-The decomposition of the graph in maximally dense groups of nodes is implemented and used to assess the quality of the embeddings in a structural way. See module *validation* and the comments on the embedding of the *Orkut* graph where we can use the community data provided with the graph to analyze the behaviour of embedded edge lengths.  
-
-In particular it is shown that :
-  - embedding of edges internal to a community are consistently smaller than embedded edges crossing a block frontier. 
-  - The transition probabilities of edge from one block to another are similar (low kullback divergence) in the original graph and in the embedded graph.
- 
-  See results in [orkut.md](./orkut.md) and examples directory together with a small Rust notebook in directory [Notebooks](./Notebooks/orkutrs.ipynb)
-
 ## Validation
 
 Validation of embeddings is assessed via standard Auc with random deletion of edges. See documentation in the *link* module and *embed* binary.
 We give also a variation based on centric quality assessment as explained at [cauc](http://github.com/jean-pierreBoth/linkauc)
 ## Some data sets
 
-### Without labels
-
 Small datasets are given in the Data subdirectory (with 7z compression) to run tests.  
 Larger datasets can be downloaded from the SNAP data collections <https://snap.stanford.edu/data>
 
@@ -132,13 +186,11 @@ A preliminary of node centric quality estimation is provided in the validation m
 - The munmun_twitter_social graph shows that treating a directed graph as an undirected graph give significantly different results in terms of link prediction AUC.
 
 
-
-
 ## Generalized Svd
 
 An implementation of Generalized Svd comes as a by-product in module [gsvd](./src/atp/gsvd.rs).
 
-## Installation and Usage
+## Detailed Installation and Usage
 
 ### Installation
 
@@ -161,21 +213,21 @@ so to the required dimension to get a valid embedding in $R^{n}$.
 - The *embed* module takes embedding and possibly validation commands (link prediction task) in one directive.  
 The general syntax is :
 
-    embed file_description [validation_command --validation_arguments] sketching mode  --embedding_arguments  
+    graphembed file_description [validation_command --validation_arguments] sketching mode  --embedding_arguments  
     for example:  
 
   For a symetric graph we get:
 
 - just embedding:
-        embed --csv ./Data/Graphs/Orkut/com-orkut.ungraph.txt --symetric  sketching --decay 0.2  --dim 200 --nbiter 
+        graphembed --csv ./Data/Graphs/Orkut/com-orkut.ungraph.txt --symetric  sketching --decay 0.2  --dim 200 --nbiter 
 
 - embedding and validation:
  
-        embed --csv ./Data/Graphs/Orkut/com-orkut.ungraph.txt  --symetric  validation --nbpass 5 --skip 0.15 sketching --decay 0.2  --dim 200 --nbiter 5
+        graphembed --csv ./Data/Graphs/Orkut/com-orkut.ungraph.txt  --symetric  validation --nbpass 5 --skip 0.15 sketching --decay 0.2  --dim 200 --nbiter 5
 
 For an asymetric graph we get 
 
-       embed --csv ./Data/Graphs/asymetric.csv  validation --nbpass 5 --skip 0.15 sketching --decay 0.2  --dim 200 --nbiter 5 
+       graphembed --csv ./Data/Graphs/asymetric.csv  validation --nbpass 5 --skip 0.15 sketching --decay 0.2  --dim 200 --nbiter 5 
 
 
     More details can be found in docs of the embed module. Use cargo doc --no-dep --bin embed (and cargo doc --no-dep) as usual.

graphembed_rs-0.1.2.dist-info/RECORD

@@ -0,0 +1,7 @@
+graphembed_rs-0.1.2.dist-info/METADATA,sha256=WEvYJP6qVNPFlFxYzzq-n3XX6L__a2ukxgYkVt_WnYo,11532
+graphembed_rs-0.1.2.dist-info/WHEEL,sha256=pKyTkFbTEakJa6xy_GKYIZO6TnrTnUxcBEc6JAhr_7o,104
+graphembed_rs-0.1.2.dist-info/licenses/LICENSE-MIT,sha256=ndZ12D28O4UkfOeoa6HP9E7IKyYG4iH79iQ6WiLs9bc,1077
+graphembed_rs/load_utils.py,sha256=fXd4-5OxdymSkb4kx39XeW_P3nEDdAuQ7Oe1D6PKDWE,2719
+graphembed_rs/__init__.py,sha256=s8WorNcMnn15CfjctNaLnXGnwQ9RlPyCR0WJQxepTTc,179
+graphembed_rs/graphembed_rs.cpython-310-darwin.so,sha256=tsLRSTexzC6Y6ZXPPOr640KrEbGEAt3O_2nznzQ1R-k,5266032
+graphembed_rs-0.1.2.dist-info/RECORD,,

graphembed_rs-0.1.2.dist-info/licenses/LICENSE-MIT

@@ -0,0 +1,25 @@
+Copyright (c) 2022 jean-pierre.both and Jianshu Zhao
+
+Permission is hereby granted, free of charge, to any
+person obtaining a copy of this software and associated
+documentation files (the "Software"), to deal in the
+Software without restriction, including without
+limitation the rights to use, copy, modify, merge,
+publish, distribute, sublicense, and/or sell copies of
+the Software, and to permit persons to whom the Software
+is furnished to do so, subject to the following
+conditions:
+
+The above copyright notice and this permission notice
+shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF
+ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
+TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
+PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT
+SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR
+IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+DEALINGS IN THE SOFTWARE.

graphembed/__init__.py

@@ -1,5 +0,0 @@
-from .graphembed import *
-
-__doc__ = graphembed.__doc__
-if hasattr(graphembed, "__all__"):
-    __all__ = graphembed.__all__

graphembed/__init__.pyi

@@ -1,79 +0,0 @@
-from typing import Optional
-
-# ----------  Embedding  ----------
-def embed_hope_rank(
-    csv: str,
-    symetric: bool,
-    target_rank: int,
-    nbiter: int,
-    output: Optional[str] = None,
-) -> None: ...
-def embed_hope_precision(
-    csv: str,
-    symetric: bool,
-    epsil: float,
-    maxrank: int,
-    blockiter: int,
-    output: Optional[str] = None,
-) -> None: ...
-def embed_sketching(
-    csv: str,
-    symetric: bool,
-    decay: float,
-    dim: int,
-    nbiter: int,
-    output: Optional[str] = None,
-) -> None: ...
-
-# ----------  Validation (returns mean AUC) ----------
-def validate_hope_rank(
-    csv: str,
-    symetric: bool,
-    target_rank: int,
-    nbiter: int,
-    nbpass: int = 10,
-    skip_frac: float = 0.1,
-    centric: bool = False,
-) -> float: ...
-def validate_hope_precision(
-    csv: str,
-    symetric: bool,
-    epsil: float,
-    maxrank: int,
-    blockiter: int,
-    nbpass: int = 10,
-    skip_frac: float = 0.1,
-    centric: bool = False,
-) -> float: ...
-def validate_sketching(
-    csv: str,
-    symetric: bool,
-    decay: float,
-    dim: int,
-    nbiter: int,
-    nbpass: int = 10,
-    skip_frac: float = 0.1,
-    centric: bool = False,
-) -> float: ...
-
-# ---------- VCMPR (precision/recall curves) ----------
-def estimate_vcmpr_hope_rank(
-    csv: str,
-    symetric: bool,
-    target_rank: int,
-    nbiter: int,
-    nbpass: int = 2,
-    topk: int = 10,
-    skip_frac: float = 0.1,
-) -> None: ...
-def estimate_vcmpr_sketching(
-    csv: str,
-    symetric: bool,
-    decay: float,
-    dim: int,
-    nbiter: int,
-    nbpass: int = 2,
-    topk: int = 10,
-    skip_frac: float = 0.1,
-) -> None: ...
-

graphembed_rs-0.1.0.dist-info/RECORD

@@ -1,7 +0,0 @@
-graphembed_rs-0.1.0.dist-info/METADATA,sha256=1bBA8fy75z8I6YGAsPAMoB-67zpAHW66cHywSb-hPj8,9901
-graphembed_rs-0.1.0.dist-info/WHEEL,sha256=pKyTkFbTEakJa6xy_GKYIZO6TnrTnUxcBEc6JAhr_7o,104
-graphembed/__init__.py,sha256=RCcLraveWf-myTsDQGePMYq-scNNfz-3Mv1baSbgAmM,123
-graphembed/__init__.pyi,sha256=3_KBFG4g9akylo32CHlm9bZStcLwxIY2X4si21ilD3w,1626
-graphembed/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-graphembed/graphembed.cpython-310-darwin.so,sha256=rpwaeaP3bH7t93yRz2edb4HD8CDEgu7bkEMFO3waTGU,5158592
-graphembed_rs-0.1.0.dist-info/RECORD,,